README

This package contains all the common functions used to run the Advanced Analytics algorithms. We provide a brief explanation of each of the modules and the corresponding functions:

• athena_data_provider contains the class AthenaDataProvider, which is initialized by giving the parameters

  • aws_access_key_id
  • aws_secret_access_key
  • s3_staging_dir
  • region_name.

  The function read_query returns a DataFrame with the information requested by the query passed as argument.

• logger contains the class Logger, which logs important messages and prints them to the terminal or CloudWatch in AWS.

• metrics contains the class Metrics, which creates the metrics and pushes them to Datadog.

• s3_client contains the class S3Client, that is initialized using boto3.client and boto3.resource and the environment variable RESULT_BUCKET.

  The function upload_files uploads the DataFrame returned by the algorithm (output_df) to the given path (output_path) in AWS, for the chosen variant (variant – typically production or the name of the experiment, if testing new features).

If changes are made to this package, it has to be updated in PyPI by doing:

1. Update the version number in the file pyproject.toml under the field version.
2. In the terminal, run the command python3 -m build. This will create two new files in the directory dist.
3. Upload the package by typing twine upload dist/* in the terminal. Since the account is protected by a two-factor authentication we have to use a token, meaning that you must
   • set your username to __token__ and
   • set your password to the token value, including the pypi- prefix.
4. Check that the package has been correcly updated in https://pypi.org/project/wag-advanced-analytics-utils/.

To use these modules in a Python script do the following:

• start by installing the package by running

  pip install wag_advanced_analytics_utils
  

  or

  pipenv install wag_advanced_analytics_utils
  

• import the desired class by typing

  from wag_advanced_analytics_utils.{module_name} import class_name
  

  where {module_name} is one of the modules of the package and class_name is (one of) the class contained in the module.